ADSM-L

Re: Max Processes

2003-03-25 12:15:36
Subject: Re: Max Processes
From: Tab Trepagnier <Tab.Trepagnier AT LAITRAM DOT COM>
To: ADSM-L AT VM.MARIST DOT EDU
Date: Tue, 25 Mar 2003 11:06:53 -0600
Stanfield,

I backup two disk pools and two four-drive libraries (a 3575 and a 3583)
to a four-drive DLT library.  Here's what I do.

1) Launch backup of "folderdisk" to DLT - 1 process; this is where I store
directories.  Wait 3 minutes, then...
2) Launch backup of disk pool upstream of 3583 to DLT - 1 process; this
holds last night's files smaller than 25 MB.  Wait 3 minutes, then...
3) Launch backup of 3575 to DLT - 2 processes; this holds backups from
small clients.  Wait 3 minutes, then...
4) Launch backup of 3583 to DLT - 3 processes; this holds backups from
large clients.

Steps 1-3 grab the four drives in the DLT, so the 3583 has to wait for DLT
mount points.  Because the first two are disk pools they complete very
quickly and free up their associated target drives.  As each drive comes
available, one of the three processes from the 3583 takes it and proceeds
with its backup.  Usually, the 3575 finishes before the 3583 (order of
magnitude difference on quantity of new data) so the third process from
the 3583 will grab one of the freed drives.

It works well, keeps the SCSI channels filled, and suffers very little
throughput loss during "bubbles" when a library is swapping source tapes.

Tab Trepagnier
TSM Administrator
Laitram LLC










Stanfield Alejandro <AStanfie AT PECOM DOT COM>
Sent by: "ADSM: Dist Stor Manager" <ADSM-L AT VM.MARIST DOT EDU>
03/25/2003 10:44 AM
Please respond to "ADSM: Dist Stor Manager"


        To:     ADSM-L AT VM.MARIST DOT EDU
        cc:
        Subject:        Max Processes


When doing BACKUP STGPOOL has anyone experimented with setting the
MaxProcess to a number greater than the available number of drives at that
moment hoping to use more drives after they become available?
We have several stgpools which we start backing up in parallel but, as
expected, some jobs randomly finish earlier than others, that leaves some
drives idle which could be used for other backup stgpool commands already
in
progress, this I presume should improve throughput.
Has anyone tried this?
Or do you just sequence all backup stgpools making each use the maximum
number of drives?

> Regards
> Alex
>

<Prev in Thread] Current Thread [Next in Thread>
  • Max Processes, Stanfield Alejandro
    • Re: Max Processes, Tab Trepagnier <=